The exocyst complex is required for the trafficking and delivery of KCa3.1 to the basolateral membrane of polarized epithelia

Abstract

Control of the movement of ions and water across epithelia is essential for homeostasis. Changing the number or activity of ion channels at the plasma membrane is a significant regulator of epithelial transport. In polarized epithelia, the intermediate-conductance calcium-activated potassium channel, KCa3.1 is delivered to the basolateral membrane where it generates and maintains the electrochemical gradients required for epithelial transport. The mechanisms that control the delivery of KCa3.1 to the basolateral membrane are still emerging. Herein, we investigated the role of the highly conserved tethering complex exocyst. In epithelia, exocyst is involved in the tethering of post-Golgi secretory vesicles with the basolateral membrane, which is required before membrane fusion. In our Fisher rat thyroid cell line that stably expresses KCa3.1, siRNA knockdown of either of the exocyst subunits Sec3, Sec6, or Sec8 significantly decreased KCa3.1-specific current. In addition, knockdown of exocyst complex subunits significantly reduced the basolateral membrane protein level of KCa3.1. Finally, co-immunoprecipitation experiments suggest associations between Sec6 and KCa3.1, but not between Sec8 and KCa3.1. Collectively, based on these data and our previous studies, we suggest that components of exocyst complex are crucially important in the tethering of KCa3.1 to the basolateral membrane. After which, Soluble N-ethylmaleimide-sensitive factor (SNF) Attachment Receptors (SNARE) proteins aid in the insertion of KCa3.1-containing vesicles into the basolateral membrane of polarized epithelia.

NEW & NOTEWORTHY Our Ussing chamber and immunoblot experiments demonstrate that when subunits of the exocyst complex were transiently knocked down, this significantly reduced the basolateral population and functional expression of KCa3.1. These data suggest, combined with our protein association experiments, that the exocyst complex regulates the tethering of KCa3.1-containing vesicles to the basolateral membrane prior to the SNARE-dependent insertion of channels into the basolateral membrane of epithelial cells.

INTRODUCTION

Control of the movement of ions, solutes, and water across epithelia is essential for maintaining homeostasis. In polarized epithelia, the intermediate-conductance Ca²⁺-activated K⁺ channel KCa3.1 (also called IK1 and SK4) is delivered to the basolateral membrane where the efflux of K⁺ maintains the electrochemical driving force required for luminal Cl⁻ secretion in tissues such as the pulmonary airway or intestine (1–6) and Na⁺ reabsorption in the colon (7, 8). In addition, KCa3.1 has various other functions in other tissues: including regulating vascular tone in endothelial cells (9, 10), and controlling erythrocyte volume (11, 12).

Critical for the function of KCa3.1 in epithelia is the maintenance of an appropriate number of channels at the basolateral membrane and regulating the open probability (P_o). The former requires precisely balancing the synthesis and delivery of channels to the basolateral membrane and subsequent endocytosis and degradation. Although the modulation of KCa3.1 P_o activity by kinases (6, 13, 14) and pharmacological means has been the subject of considerable research [for reviews, see Christophersen and coworkers (15, 16); Wulff and coworkers (17); Devor et al. (6)], the exact molecular mechanisms that regulate the trafficking of KCa3.1 are still poorly understood. However, we and others have begun to elucidate these pathways. Summarily, both the N- and C-terminal domains of KCa3.1 contain leucine zippers that are required for both folding and assembly of KCa3.1 into tetramers (18–20). Once KCa3.1 has been correctly assembled and folded it exits the endoplasmic reticulum (ER) and is trafficked to the Golgi in a Rab1-GTPase-dependent pathway (21). Delivery of KCa3.1 from the trans-Golgi network (TGN) to the basolateral membrane is dependent on Rab8, and independent of recycling endosomes and the µ1B subunit of the epithelial-specific adaptor protein AP-1B, a classical regulator of basolateral trafficking (21). Previously, we established that the microtubule network, cytoskeleton, and the motor protein myosin-Vc are essential for the delivery of KCa3.1-containing vesicles to the basolateral membrane (22). Pharmacological disruption to the microtubule network and cytoskeleton, or siRNA-knockdown of myosin significantly decreased KCa3.1 trafficking in polarized epithelia (22). Finally, we recently identified that the insertion of KCa3.1 into the basolateral membrane requires interactions between the Soluble N-ethylmaleimide-sensitive factor (SNF) Attachment Receptors (SNARE) proteins syntaxin-4 (STX-4), VAMP3, and SNAP23 (23).

Prior to SNARE-mediated fusion, however, secretory vesicles must be tethered to the plasma membrane. The exocyst complex is an eight-protein complex that regulates protein trafficking by tethering postsecretory vesicles with the plasma membranes before vesicle fusion. Originally identified in budding Saccharomyces cerevisiae (24), exocyst is highly conserved in higher eukaryotes (25, 26). The mammalian form of exocyst comprises eight subunits: Sec3, Sec5, Sec6, Sec8, Sec10, Sec15, Exo70, and Exo84 (also known as EXOC1–EXOC8, respectively). Exocyst is a member of the complexes associated with tethering containing helical rods (CATCHR) family of proteins. Prior to the fusion of a vesicle with the target membrane, the two membranes must be in close proximity to each other. By interacting with specific proteins and lipids on both membranes, tethering complexes can bring the membrane close enough for SNARE-medicated fusion (27). Briefly, exocyst is recruited to secretory vesicles by Rab-GTPases (28), next, the Pleckstrin homology (PH) domain in the N-terminal region of Sec3 binds to the phospholipid phosphatidylinositol (4, 5)-bisphosphate [PI(4,5)P₂] in the plasma membrane (29–31). Concurrently, Exo70 binds to PI(4,5)P₂ through a series of conserved basic residues in its C-terminal domain (32, 33). This interaction brings the vesicle in contact with the plasma membrane to initiate SNARE-medicated fusion. SNARE proteins on both the vesicle and in the plasma membrane form a trans-SNARE complex and the two membranes are “zipped” together (34). In polarized epithelial cells, exocyst is generally restricted to tethering vesicles with the basolateral membrane (35, 36). Here, exocyst has a critical role in the delivery of membrane junctional proteins, such as E-cadherin and members of the claudin family, to the basolateral membrane (37–39). However, exocyst’s role in regulating the trafficking of ion channels and transporters is still an evolving area of research.

In this report, we demonstrate that the exocyst complex is required for the targeting and trafficking of KCa3.1 to the basolateral membrane of polarized epithelia. When protein levels of the subunits Sec3, Sec6, or Sec8 were reduced by siRNA, the basolateral membrane population of KCa3.1 was significantly reduced. In addition, depletion of the exocyst subunits significantly reduced the functional expression of KCa3.1, as measured by KCa3.1-specific transepithelial current. Finally, co-immunoprecipitation (Co-IP) experiments identified interactions between KCa3.1 and Sec6, but not Sec8. Collectively, these data suggest that the exocyst complex is required for the tethering of KCa3.1 at the basolateral membrane. Given the important role of KCa3.1 in maintaining electrochemical gradients for ion and solute transport in epithelial, understanding the trafficking pathways of KCa3.1 is critical for the development of therapeutic interventions.

MATERIALS AND METHODS

Molecular Biology—Internal Biotinylation of KCa3.1

The development of biotin ligase acceptor peptide (BLAP) tagged KCa3.1 has been described in detail previously (22, 40–43). Succinctly, the BLAP sequence (GLNDIFEAQKIEWHE) was inserted into the second extracellular loop between the S3 and S4 membrane-spanning domains. cDNA for both KCa3.1-BLAP and the biotin ligase BirA containing the endoplasmic reticulum (ER) retention motif KDEL [a gift from Prof Alice Y. Ting; M.I.T., Cambridge, MA (44)] were subcloned into the pBudCE4.1 plasmid (Invitrogen). A stable cell was established that stably expresses this plasmid (described in Ref. 22); consequently, BirA-KDEL biotinylates KCa3.1 before the channel exiting the ER (44). We have previously reported that the BLAP sequence does not alter the biophysical or pharmacological properties of the channel (45). Furthermore, in Fisher rat thyroid (FRT) cells, we have demonstrated that KCa3.1 is exclusively trafficked to the basolateral membrane (21, 22, 43).

Cell Culture and Transient Transfections

Fisher rat thyroid (FRT) cells (RRID:CVCL_A61E) were maintained in Coon’s modification F12 nutrient mixture [Sigma-Aldrich, F6636, New Zealand (NZ)] and HEK-293 cells (RRID:CVCL_0045) were maintained in high-glucose DMEM (Gibco, Life Technologies, NZ). All media were supplemented with 10% FBS (Cytiva, NZ), 10 U/mL penicillin, and 100 µg/mL streptomycin (Gibco, Life Technologies, NZ). For routine culturing, cells were grown in 25-cm² tissue culture flasks (Nunc, Thermo Fisher Scientific, NZ). Cell lines were maintained at 37°C in the presence of 5% CO₂/95% air in a humidified cell culture incubator. Stable expression of KCa3.1-BLAP and BirA-KDEL was achieved with the antibiotic Zeocin (850 μg/mL; Gibco, Life Technologies, NZ).

The Lipofectamine 3000 transfection kit (Invitrogen, Life Technologies, NZ) was used to transfect siRNA or plasmid DNA following the manufacturer’s protocol. In this project, the following siRNAs were used against Sec6 (Exoc3) 5′-AUAAUAGCCGUCGUCGUCUGCUUCC-3′, Sec8 (Exoc4) 5′-GAUGCAUGACUUGAGUGCUAUUCA-3′, Sec3 (Exoc1) 5′-GCCCUGAGUUUGAUUUGC-3′, and the Stealth RNAi negative control siRNA (all from Invitrogen). Concentration of 20 pmol·µL⁻¹ was used for all siRNA transfections.

Biotinylation and Streptavidin Labeling of KCa3.1-BLAP Channels

Our protocol for streptavidin labeling of basolateral KCa3.1-BLAP channels has been detailed previously (22, 43). In short, FRT-KCa3.1-BLAP cells were cultured on 24-mm Transwell permeable supports (Corning Inc., via In Vitro Technologies Ltd., NZ) until reaching a confluent epithelial monolayer (∼72 h postseeding). The monolayer was first washed with ice-cold PBS, followed by two washes with 1% BSA in PBS. Next, the basolateral membrane was incubated with 10 µg/mL streptavidin (Thermo Fisher Scientific, NZ) for 30 min at 4°C. Following incubation, the monolayer was washed extensively with 1% BSA to quench unbound streptavidin. Proteins were extracted using lysis buffer [150 mmol·L⁻¹ NaCl, 50 mmol·L⁻¹ HEPES (pH 7.4), 1 mmol·L⁻¹ EDTA, 1% Triton X-100, and 1× cOmplete protease inhibitor cocktail (Roche)]. Lysate protein concentration was determined with the DC assay kit (Bio-Rad).

Immunoblotting

Immunoblotting was performed using standard protocols as described previously (18, 19, 21, 22, 43). In short, 30 µg of protein was loaded per lane and was separated on hand-cast 8% SDS-PAGE gels and transferred to PVDF membranes (Roche Biochemicals, via Sigma-Aldrich, NZ) using a Trans-Blot Turbo semi-dry transfer system (Bio-Rad, Hercules, CA) at 1.3 A for 35 min. GAPDH served as a loading control. Membranes were blocked in 5% nonfat milk powder (Fonterra, NZ) in Tris-buffered saline wash buffer [(TBS-T); 20 mmol·L⁻¹ Tris-base (pH 7.5), 150 mmol·L⁻¹ NaCl, 1% Tween 20] for 1–2 h at room temperature. Membranes were incubated overnight at 4°C with primary antibodies against Claudin-1 (1:1,000, Thermo Fisher Scientific, 37-4800, RRID:AB_2533323), GAPDH (1:5,000, Sigma-Aldrich, G9545, RRID:AB_796208), E-cadherin (1:4,000, Cell Signaling Technology, 3195, RRID:AB_2291471), hemagglutinin epitope tag (HA; 1:1,000, Sigma-Aldrich, H6908, RRID:AB_260070), Sec3 (1:2,000, ProteinTech, 11690-1-AP, RRID:AB_2231500), Sec6 (1:5,000, Abcam, AB156568, RRID:AB_2924910), Sec8 (1:1,000, Santa Cruz Biotechnology, SC-136234, RRID:AB_2101573), streptavidin (1:2,000, GeneScript, A00621-10, RRID:AB_2924911), SNAP23 (1:1,000, Santa Cruz Biotechnology, SC-166244, RRID:AB_2286322), STX-4 (1:1,000, Merck Millipore, 574788, RRID:AB_212525), VAMP3 (1:1,000, Sapphire Bioscience, Australia, C144617, RRID:AB_10968088). Membranes were extensively washed in TBS-T buffer before being incubated in secondary antibody, either goat anti-rabbit (1:10,000, Sigma-Aldrich, A6154, RRID:AB_258284), or goat anti-mouse (1:10,000, Sigma-Aldrich, A4416, RRID:AB_258167) for 1 h at room temperature. Protein bands were visualized by enhanced chemiluminescence (Bio-Rad Clarity) and detected using either autoradiographic film (Carestream, via Radiographic Supplies Ltd.) or a Bio-Rad ChemiDoc MP imaging system (Model No. 17001402) running Image Lab Touch (v.2.4). Normalization of the immunoblot data were as follows: the experimental band intensity was divided by the control band intensity followed by division of this ratio by a similar ratio for the GAPDH band intensities. This normalization process makes all of the bars representing the siControl densities exactly equal to 1 by definition. Protein molecular weight was estimated against the BenchMark prestained protein standard (Invitrogen, 1645050) or Precision Plus prestained standard (Bio-Rad, 1610374). When probing PVDF membranes with multiple primary antibodies, the membrane was stripped of primary and secondary antibodies before re-probing. Blots were stripped with Restore Western Blot Stripping Buffer (Thermo Scientific, 21059) at 37°C for 15 min following the manufacturer’s instructions.

Ussing Chamber Electrophysiology

Ussing chamber electrophysiology experiments were conducted as previously described (22, 23, 43). To measure transepithelial current (I_t), FRT-KCa3.1-BLAP cells were cultured on 12-mm Snapwell filters (Corning Costar, via In Vitro Technologies, NZ) until they formed a confluent monolayer (∼72 h postseeding) which was then mounted into a modified Ussing chamber and clamped at 0 mV. To facilitate transepithelial transport of K⁺, a K⁺ gradient was applied across the cell monolayer: the mucosal chamber was filled with 5 mL of high K⁺ concentration Ringer’s solution [containing (in mmol·L⁻¹) 145 K-gluconate, 10 HEPES, 1 MgCl₂, 4 CaCl₂, 10 d-glucose, pH adjusted to 7.4 with KOH]; the serosal chamber was filled with 5 mL low K⁺ concentration Ringer’s solution [containing (in mmol·L⁻¹) 140 Na-gluconate, 5 K-gluconate, 10 HEPES, 1 MgCl₂, 4 CaCl₂, 10 d-glucose, pH adjusted to 7.4 with NaOH]. Note that the CaCl₂ concentration was increased from the typical 1.2 mmol·L⁻¹ to compensate for Ca²⁺ buffering by gluconate anions (46). The Ussing chamber system consisted of an eight chamber EasyMount Ussing chamber connected to an eight-channel voltage-current clamp (both from Physiologic Instruments Inc., Reno, NV) and a DI-720 data acquisition system (DataQ Instruments, Akron, OH). Data were recorded using the Acquire and Analyze program (v.2.3, Physiologic Instruments). Transepithelial resistance (R_t) was calculated by applying a 5-mV pulse every 120 s. FRT monolayers typically form a tight epithelium; accordingly, the monolayer was considered to have sufficient integrity if the R_t was ≥ 300 Ω·cm². Monolayers that did not meet this threshold were excluded from any analysis. To measure the functional expression of KCa3.1, the change in the current before and after the addition of the KCa3.1 positive-gating modulator 1-EBIO (1) to both the mucosal and serosal chambers (final concentration 100 µmol·L⁻¹) was measured. To confirm the specificity of this change in current, the KCa3.1 blocker clotrimazole (CLT) (2) was added to both the mucosal and serosal chambers (final concentration 10 µmol·L⁻¹). We previously demonstrated that FRT cells not transfected with KCa3.1 do not respond to 1-EBIO or CLT indicating FRT epithelia lack endogenous KCa3.1 (43).

Co-Immunoprecipitation Experiments

For Co-IP experiments, HEK-293 cells were cultured on 100-mm dishes (Greiner Bio-One via Lab Supply Ltd.) and transfected with 3 µg plasmid encoding HA-tagged KCa3.1 using Lipofectamine 3000. Alternatively, FRT-KCa3.1-BLAP cultured on 24-mm Transwell filters were used. The protocol for both cells was identical. Cells were lysed 48 h post-transfection in lysis buffer, the protein concentration was determined with the DC protein assay kit, and 50 µg of protein was reserved for the “input.” Equal amounts of the crude lysates were incubated with either 2 µL of anti-Sec6, anti-Sec8, anti-HA, or goat anti-rabbit IgG (negative control) antibody for 3–4 h at 4°C with end-over-end mixing. Next, 50 µL of PBS equilibrated protein G Sepharose beads (Sigma-Aldrich, P3296) were added to the lysate and incubated for a further 1–2 h at 4°C with end-over-end mixing. The beads were then extensively washed with TBS + 1% Triton X-100. Finally, proteins were eluted from the beads by incubating in SDS sample buffer for 5 min at 95°C. Immunoprecipitates were resolved by SDS-PAGE and analyzed by immunoblotting as described earlier.

Chemicals

All chemicals were purchased from Sigma-Aldrich unless otherwise stated. Both 1-EBIO and clotrimazole were solubilized in DMSO.

Statistical Analysis and Densitometry

Ussing chamber data were exported to Microsoft Excel for preliminary analysis to calculate 1-EBIO-sensitive current. All statistical analysis was performed using GraphPad Prism (GraphPad Software, v.9.3.2). An unpaired t test followed by a Welch’s post-test was used to compare between two groups. For multiple comparisons, a one-way ANOVA with Tukey’s post-test was used. All data are presented as means ± standard deviation (SD), where n indicates the number of experiments in different passages of cells; all experiments were performed on a minimum of three cell passages. A P ≤ 0.05 was regarded as statistically significant. Semiquantitative densitometry was used to analyze immunoblot experiments. In short, relative band density was determined with Image J (NIH, v1.53) and density was normalized to the control condition. A one-sample t test followed by a Welch’s post-test was used to compare between groups.

RESULTS

Sec6 and Sec8 Are Required for Delivery of KCa3.1 to the Basolateral Membrane

Based on the previous studies detailed in introduction, we hypothesized that exocyst was required for the tethering of KCa3.1-containing postsecretory vesicles at the basolateral membrane. To investigate this hypothesis, we carried out siRNA-mediated knockdown of Sec proteins and determined the expression of KCa3.1 at the basolateral membrane. We initially focused on the exocyst subunits Sec6 and Sec8, as these proteins form a hetero-dimer, localize to the basolateral membrane, and have previously been reported to regulate the targeting and trafficking of basolateral proteins (35).

As seen in the immunoblot in Fig. 1A (top), transfection of Sec6 siRNA significantly reduced Sec6 protein levels by 53 ± 18% (Fig. 1C, n = 6, P ≤ 0.001). In FRT epithelia where Sec6 had been knocked down, basolateral membrane KCa3.1 levels were significantly decreased (Fig. 1A, middle), averaging 69 ± 11% compared with control transfected cells (Fig. 1E, n = 6, P ≤ 0.0001). As can be seen in Fig. 1B, FRT cells transiently transfected with Sec8 siRNA exhibited a significant reduction of Sec8 protein levels by 60 ± 27% (Fig. 1D, n = 7, P ≤ 0.01) compared with control cells. Similarly, in FRT epithelia where Sec8 had been knocked down, the basolateral membrane KCa3.1 level was significantly decreased (Fig. 1B, middle) averaging 80 ± 14% compared with control cells (Fig. 1F, n = 7, P ≤ 0.0001). These immunoblot experiments suggest that knockdown of either Sec6 or Sec8 impairs the delivery of KCa3.1 to the basolateral membrane, indicating Sec6 and Sec8 are required for tethering KCa3.1-containing vesicles at the basolateral membrane.

Figure 1.

Knockdown of the exocyst subunits Sec6 or Sec8 decreases the basolateral population of intermediate-conductance calcium-activated potassium channel (KCa3.1) in Fisher rat thyroid (FRT) epithelia. FRT-KCa3.1-biotin ligase acceptor peptide (BLAP) cells were transiently transfected with either a scrambled control siRNA, Sec6 siRNA, or Sec8 siRNA and were cultured on Transwell permeable supports to form a confluent monolayer. Seventy-two hours postseeding, KCa3.1 channels were labeled with streptavidin before FRT-KCa3.1-BLAP epithelia were lysed and analyzed by immunoblotting. KCa3.1 was detected with an anti-streptavidin antibody; GAPDH was used as the loading control. A: knockdown of Sec6 in FRT-KCa3.1-BLAP cells. Transfection of Sec6 siRNA (top) significantly reduced endogenous Sec6 levels (top) and basolateral levels of KCa3.1 (middle). B: knockdown of Sec8 in FRT-KCa3.1-BLAP cells. Transfection of Sec8 siRNA (top) significantly reduced endogenous Sec8 levels (top) and reduced basolateral KCa3.1 (middle). C and D: relative band density was determined semiquantitively. The graphs plot relative Sec6 density (C) and Sec8 (D) normalized to GAPDH. Data are presented as means ± SD, n = 6 or 7, one-sample t test, **P ≤ 0.01; ***P ≤ 0.001. E and F: the graphs plot relative basolateral membrane KCa3.1 density normalized to GAPDH. Data are presented as means ± SD, reduced Sec6 (E), n = 6, one-sample t test, ****P < 0.0001. F: Sec 8, n = 7 one-sample t test, ****P ≤ 0.0001.

Knockdown of Exocyst Subunits Sec6 and Sec8 Alters KCa3.1 Current

Based on our immunoblot experiments, we hypothesized that knockdown of Sec6 or Sec8 would reduce the functional expression of KCa3.1 at the basolateral membrane. Employing Ussing chamber experiments, we induce KCa3.1 current in FRT epithelia with the KCa3.1 potentiator 1-EBIO (1) and confirm this increase in current is carried by KCa3.1 channels by application of the KCa3.1 blocker clotrimazole (2) as we have reported (22, 43). As shown in the representative current trace (Fig. 2A), when Sec6 levels were reduced (dotted line), there was a significant reduction in KCa3.1-specific current by 85 ± 10% (Fig. 2B, n = 10, P ≤ 0.01) compared with control transfected cells. Likewise, as seen in Fig. 2C, when levels of Sec8 were reduced by siRNA (dotted line), there was a significant reduction of KCa3.1 specific current by 78 ± 21% (Fig. 2D, n = 12, P ≤ 0.01) compared with control cells. These data corroborate our immunoblot data, suggesting that when Sec6 or Sec8 are knocked down fewer KCa3.1 channels are present at the basolateral membrane.

Figure 2.

Knockdown of the exocyst subunits Sec6 or Sec8 decreases intermediate-conductance calcium-activated potassium channel (KCa3.1) current in Fisher rat thyroid (FRT) epithelia. FRT-KCa3.1-biotin ligase acceptor peptide (BLAP) cells transiently transfected with either scrambled control siRNA (solid line), Sec6 siRNA (dashed line), or Sec8 siRNA (dotted line) and were cultured on Snapwell filter for 72 h. Confluent FRT-KCa3.1-BLAP monolayers were mounted into a modified Ussing chamber and clamped at 0 mV. A and C: representative current traces. The addition of the KCa3.1 positive-gating modulator 1-EBIO (100 µM) and the KCa3.1-specific blocker clotrimazole (CLT; 10 µM) are indicated as arrows and capped arrows, respectively. B: when Sec6 levels were reduced, there was a significant reduction in KCa3.1-specific current by 85 ± 10% (n = 10, P ≤ 0.01, unpaired t test) compared with control transfected cells. D: when levels of Sec8 were reduced by siRNA transfection, there was a significantly reduction of KCa3.1 specific current by 78 ± 21% (n = 12, P ≤ 0.01, unpaired t test) compared with control transfected cells. E: transepithelial resistance (R_t) of FRT-KCa3.1-BLAP epithelia. Data are shown as means ± SD. Control transfected FRT epithelia had a mean R_t of 1,020 ± 587 Ω·cm², siSec6 transfected FRT epithelia had a mean R_t of 664.2 ± 229.4 Ω·cm², and siSec8 transfected FRT epithelia had a mean R_t of 977.1 ± 513.7 Ω·cm². One-way ANOVA, ns indicates P ≥ 0.05, n = 10–12. The dotted line indicates the threshold resistance (300 Ω·cm²). **P < 0.01.

It has previously been reported that knockdown of exocyst proteins can result in a reduction of membrane junctional proteins that could lead to reduced integrity of the basolateral membrane of the epithelial cells (37, 38). To directly assess this possibility, we compared the mean R_t of FRT-KCa3.1-BLAP epithelia transfected with either Sec6 or Sec8 siRNAs to control transfected epithelia (Fig. 2E). R_t is a real-time measure of epithelial integrity as confluent tight epithelial monolayers such as FRT typically have R_t > 300 Ω·cm². Control transfected FRT epithelia had a mean R_t of 1020 ± 587 Ω·cm², siSec6 transfected FRT epithelia had a mean R_t of 664.2 ± 229.4 Ω·cm², and siSec8 transfected FRT epithelia had a mean R_t of 977.1 ± 513.7 Ω·cm². Importantly, knockdown of either Sec6 or Sec8 did not significantly alter R_t (Fig. 2E, n = 10–12, P ≥ 0.05) suggesting that the observed decrease in 1-EBIO-sensitive current was due to a decrease in functional expression of KCa3.1 and not impaired epithelial integrity.

To further investigate the possible loss of membrane integrity due to altered membrane junctional proteins, we examined the cellular level(s) of E-cadherin and claudin-1, which are key components of membrane junctions, following knockdown of Sec6 and Sec8. Therefore, using immunoblot experiments, we used the same lysates from the knockdown of Sec6 and Sec8 experiments (Fig. 1, A and B), and probed for E-cadherin and claudin-1. As seen in Fig. 3, FRT-KCa3.1 epithelia transfected with either Sec6 siRNA (Fig. 3, A and B) or Sec8 siRNA (Fig. 3, A and C) did not alter the levels of E-cadherin or claudin-1 (Fig. 3, D–G, n = 3–4, P ≥ 0.05). These results, coupled with the transepithelial resistance data (Fig. 2E), strongly suggest that the integrity of the basolateral membrane of the epithelial cells was not affected by the knockdown of Sec6 or Sec8.

Figure 3.

Knockdown of the exocyst complex proteins Sec6 or Sec8 does not alter levels of the junctional proteins E-cadherin or claudin-1. Fisher rat thyroid (FRT)-intermediate-conductance calcium-activated potassium channel (KCa3.1)-biotin ligase acceptor peptide (BLAP) cells were transiently transfected with either a scrambled control siRNA, Sec6 siRNA, or Sec8 siRNA and were cultured on Transwell permeable supports to form a confluent monolayer. Seventy-two hours postseeding, cells were lysed and analyzed by immunoblotting. A: lysates for Sec6 (B) knockdown and Sec8 knockdown (C) cells underwent immunoblotting for E-cadherin and claudin-1. D–G: the graphs plot relative band density normalized to GAPDH. Knockdown of either Sec6 or Sec8 did not significantly alter the levels of E-cadherin nor claudin-1. Data are presented as means ± SD, n = 3, one-sample t test, *P < 0.05, ns indicates P > 0.05.

The Exocyst Subunit Sec3 Regulates KCa3.1 Trafficking

Thus far we have investigated, only two of the eight exocyst proteins, Sec6 and Sec8, are regulators of KCa3.1 trafficking. To further validate our data on exocyst-dependent trafficking of KCa3.1, the role of Sec3 was studied. Sec3 contains a conserved PH domain that is able to bind to the phospholipid PI(4,5)P₂ (30, 31). In polarized epithelia, PI(4,5)P₂ is generally restricted to the basolateral membrane (47). As such, we hypothesized that since the Sec3 subunit is a key interface between the secretory vesicle and the plasma membrane it would be critical for targeting KCa3.1-containing secretory vesicles to the basolateral membrane. First, immunoblot experiments were conducted identically to those described for Sec6 and Sec8. Transfection of Sec3 siRNA significantly decreased Sec3 protein levels by 51 ± 27% (Fig. 4, A and B, n = 4, P ≤ 0.05) compared with control transfected cells. Knockdown of Sec3 (Fig. 4, A and C) significantly reduced the basolateral population of KCa3.1 (Fig. 4A, middle), averaging 64 ± 31% compared with control transfected cells (Fig. 4C, n = 4, P ≤ 0.05).

Figure 4.

Knockdown of the exocyst complex protein Sec3 reduces basolateral membrane levels and current of intermediate-conductance calcium-activated potassium channel (KCa3.1) in Fisher rat thyroid (FRT) epithelia. FRT-KCa3.1 cells transiently reverse transfected with either a scrambled control siRNA (siControl) or Sec3 siRNA (siSec3) and were cultured on Transwell filters. After the cells had formed a confluent monolayer (72 h postseeding), basolateral KCa3.1 channels were labeled with streptavidin before FRT-KCa3.1 cells were lysed and subsequently analyzed by immunoblotting to detect Sec3 and KCa3.1, using an anti-Sec3 antibody and anti-streptavidin antibody, respectively. A: representative immunoblots showing the presence of Sec3 (top) and KCa3.1 (middle) in FRT epithelia. The housekeeping protein GAPDH (bottom) was used as the loading control. B: transfection of Sec3 siRNA significantly decreased Sec3 protein levels by 51 ± 27% (n = 4, P ≤ 0.05, one-sample t test) compared with control transfected cells. Relative band density was determined semiquantitively. The graph plots relative Sec3 density normalized to GAPDH. C: knockdown of Sec3 using siRNA transfection significantly reduced the basolateral population of KCa3.1 by 64 ± 31% (n = 4, P ≤ 0.05, one-sample t test) compared with control transfected cells. The graph plots relative basolateral membrane KCa3.1 density was normalized to GAPDH. D–F: to examine the role of knockdown of Sec3 on the functional expression of KCa3.1, FRT-KCa3.1-biotin ligase acceptor peptide (BLAP) cells transiently transfected with either a scrambled control siRNA or Sec3 siRNA were cultured on Snapwell filters for 72 h. Confluent FRT-KCa3.1-BLAP monolayers were then mounted into a modified Ussing chamber and clamped at 0 mV. D: representative current traces. The addition of the KCa3.1 positive-gating modulator 1-EBIO (100 µM) and the KCa3.1-specific blocker clotrimazole (CLT; 10 µM) are indicated by arrows and capped arrows, respectively. E: knockdown of Sec3 significantly decreased KCa3.1 current by 92 ± 4% (n = 5, P ≤ 0.01, unpaired t test) compared with control transfected FRT-KCa3.1 cells. F: transepithelial resistance (R_t). Data are shown as means ± SD. Unpaired t test P ≥ 0.05. The dotted line indicates the threshold resistance (300 Ω·cm²). Control transfected FRT epithelia had a mean R_t of 689.1 ± 221.7 Ω·cm²; Sec3 transfected FRT epithelia had a mean R_t of 650.6 ± 292.9 Ω·cm² (n = 5 for both, P ≥ 0.05). Data are presented as means ± SD. *P ≤ 0.05, **P ≤ 0.01.

To test if this reduction in the basolateral population of KCa3.1 corresponded with a decreased functional expression of KCa3.1, we conducted Ussing chamber experiments with FRT-KCa3.1-BLAP epithelia transfected with either a control siRNA (Fig. 4D, bottom trace) or Sec3 siRNA (Fig. 4D, top trace). Knockdown of Sec3 significantly decreased KCa3.1 current by 92 ± 4% (Fig. 4E, n = 6, P ≤ 0.01) compared with control transfected FRT-KCa3.1 cells. Importantly, transfection of Sec3 siRNA did not significantly alter R_t (Fig. 4F, n = 5, P ≥ 0.05). Examining membrane integrity, control transfected FRT epithelia had a mean R_t of 689.1 ± 221.7 Ω·cm²; Sec3 transfected FRT epithelia had a mean R_t of 650.6 ± 292.9 Ω·cm². Accordingly, this confirms that the reduction in KCa3.1-specific current was due to a change in the basolateral population of KCa3.1, not a loss of epithelial integrity.

Does the Exocyst Complex Directly Associate with KCa3.1?

Based on our earlier data, and the known ability of exocyst proteins to associate with their cargo (28, 36, 48), we determined whether Sec6 or Sec8 similarly associate with KCa3.1. To test this hypothesis, we conducted Co-IP experiments. For these experiments, we used HEK293 cells, because, previously we have achieved more robust channel expression with transient transfections using these cells with HA-tagged channels (40, 45, 49). As shown in Fig. 5, when endogenous Sec6 was used as the “bait,” HA-KCa3.1 Co-IPs with Sec6 (Fig. 5A, bottom, lane 4, n = 4). However, when endogenous Sec8 was used as the “bait,” HA-KCa3.1 did not Co-IP (Fig. 5B, bottom, lane 4, n = 4). Since we did not observe an association between Sec 8 and HA-KCa3.1 in HEK cells, we conducted additional Co-IP experiments with our stably expressing KCa3.1-BLAP cell line. Using endogenous Sec8 as “bait” and blotting for streptavidin-labeled KCa3.1, again, there was no detectable association between Sec8 and KCa3.1 (Fig. 5C, lane 3, n = 3). We note that the Co-IP lanes ran slightly higher compared with the input lanes. Our in-house made gels run slightly unevenly resulting in these small differences. In addition, it might be possible that the salt concentration varies between the input and the IP that may affect protein mobility in the gel and the protein runs slightly different from the input.

Figure 5.

Protein-protein interactions between the exocyst complex and intermediate-conductance calcium-activated potassium channel (KCa3.1). HEK293 cells were transiently transfected with HA-tagged KCa3.1 plasmid and lysed 48 h post-transfection. A and B: Sec 6 associates with HA-KCa3.1, however Sec8 does not associate. Crude lysates were incubated with either anti-HA, anti-Sec6, anti-Sec8, or anti-rabbit IgG (negative control) antibodies. Immunocomplexes were precipitated with protein G Sepharose beads and analyzed by immunoblotting with anti-HA (A and B, bottom blots), anti-Sec6 (A, top blot), or anti-Sec8 (B, top blot). (n = 4 each). IgG LC, Immunoglobulin G light chain. C: Sec8 does not associate with KCa3.1 in Fisher rat thyroid (FRT) epithelia. Since we did not observe an association between Sec 8 and HA-KCa3.1 in HEK cells, we conducted additional Co-IP experiments with our stably expressing KCa3.1-biotin ligase acceptor peptide (BLAP) cell line. Before being lysed, the basolateral KCa3.1 channels were labeled with streptavidin. Lysates were incubated with either anti-Sec8 or goat anti-mouse IgG antibody (negative control) and protein-G Sepharose beads. Both the whole cell lysates (i.e., input) and immunoprecipitates were analyzed by immunoblotting. KCa3.1 was detected with an anti-streptavidin antibody and Sec8 was detected with an anti-Sec8 antibody (n = 3).

Knockdown of Exocyst Does Not Alter SNARE Protein Levels

Previously, we identified that the SNARE proteins STX-4, VAMP3, and SNAP23 are required for the insertion of KCa3.1-containing vesicles into the basolateral membrane of epithelial cells (22). Tangentially, it has been established by both Co-IP experiments and mass spectrometry analysis that exocyst subunits associate with SNARE proteins, such as SNAP23 (48, 50, 51). Accordingly, there is evidence in the literature that members of the exocyst complex promote assembly of SNARE proteins at the vesicle tethering site before SNARE-dependent insertion of vesicles into the membrane (51–55). As demonstrated earlier, knockdown of exocyst subunits decreased the KCa3.1 population at the basolateral membrane and KCa3.1 current (Figs. 1 and 2). As an additional control and to confirm that the impaired trafficking of KCa3.1 described earlier was due solely to the function of exocyst, rather than a coordinated effect involving both SNARE and exocyst proteins, we conducted additional experiments to semiquantitatively measure SNARE protein levels. We hypothesized that since tethering of secretory vesicles occurs before SNARE-mediated fusion, knockdown of Sec6 and Sec8 would not affect the cytosolic levels of the SNARE proteins. Thus, using immunoblot experiments, we used the same lysates from the knockdown of Sec6 and Sec8 experiments (Fig. 1, A and B), and probed for the SNARE proteins. As can be seen in Fig. 6, transfection of either Sec6 siRNA (Fig. 6A) or Sec8 siRNA (Fig. 6B) into FRT-KCa3.1 epithelia did not significantly alter the levels of STX-4, VAMP3, or SNAP23 (Fig. 6, C–J, n = 3, P ≥ 0.05). These data suggest that the decrease in the basolateral population of KCa3.1 in exocyst knockdown cells is due to impaired tethering of KCa3.1-containing vesicles upstream of SNARE protein function.

Figure 6.

Knockdown of exocyst does not alter Soluble N-ethylmaleimide-sensitive factor (SNF) Attachment Receptors (SNARE) protein levels. Fisher rat thyroid (FRT)-intermediate-conductance calcium-activated potassium channel (KCa3.1)-biotin ligase acceptor peptide (BLAP) cells were transiently transfected with either a scrambled control siRNA, Sec6 siRNA, or Sec8 siRNA and were cultured on Transwell permeable supports to form a confluent monolayer. Seventy-two hours postseeding, confluent epithelia were lysed and analyzed by immunoblotting. Knockdown of Sec6 (A and C) or Sec8 (B and G) did not significantly alter cytosolic levels of syntaxin-4 (STX-4), SNAP23, or VAMP3 (D–F and H–J). Data are presented as means ± SD. *P < 0.05, ns indicates P > 0.05, one-sample t test, n = 3.

DISCUSSION

In polarized epithelia, KCa3.1 is a critical regulator of ion and solute transport. Essential for these functions is the tight control of the channel activity by modulation of the P_o and the number of channels at the membrane (n). Maintaining an appropriate basolateral membrane population of channels requires balancing the synthesis and delivery of new channels to the plasma membrane, endocytosis, and degradation. The mechanisms that regulate the trafficking and delivery of KCa3.1 to the basolateral membrane is still emerging. Nevertheless, we and others have in recent years begun to elucidate these molecular mechanisms. Briefly, the anterograde trafficking of KCa3.1 is dependent on the Rab-GTPases Rab1 and Rab8 (21) and the motor protein Myosin-Vc (22). In addition, KCa3.1 is trafficked directly to the basolateral membrane independent of recycling endosomes and the µ1B subunit of the epithelial-specific adaptor protein AP-1B, a classical regulator of basolateral trafficking (21). Once KCa3.1-containing vesicles are tethered at the basolateral membrane, SNARE-mediated fusion results in the channel being inserted into the basolateral membrane (23).

Interestingly, disruption in the basolateral membrane population of KCa3.1 has been implicated in the pathological phenotype of several diseases. First, in polycystic kidney disease an increased expression of KCa3.1 increases the electrochemical driving force for Cl^– secretion—and consequently also increases the osmotic driving force for water secretion—into the cyst lumen. This accelerates the decline in kidney function (56). In contrast, in ulcerative colitis, reduced expression of KCa3.1 decreases the electrochemical gradient for ion and solute absorption in the colon thereby contributing to the diarrhea characteristic of inflammatory bowel diseases such as ulcerative colitis (57). These two pathologies highlight the importance of regulating the correct number of KCa3.1 channels at the basolateral membrane. In the years since the cloning of KCa3.1, considerable research has focused on the pursuit of compounds to alter KCa3.1 activity to treat diseases such as cystic fibrosis (58) and sickle cell anemia (59–61).

In this report, we investigated the role of the exocyst, in regulating the tethering of KCa3.1-containing vesicles with the plasma membrane before SNARE-mediated fusion of the vesicles. In mammals, the exocyst complex has been identified as a key regulator of the trafficking of various membrane-bound proteins. In polarized epithelial cells, the exocyst regulates the delivery of a range of membrane-bound proteins including Nephrin and Neph1 (62), E-cadherin (37, 38), Par3 (63), and aquaporin-2 containing vesicles (64). Whether exocyst regulates ion channel trafficking in epithelia is a much less studied area. Based on previous findings, we hypothesized that exocyst regulates the tethering of KCa3.1-containing vesicles at the basolateral membrane upstream of SNAREs, similar to how associations between SNARE and exocyst proteins regulate the trafficking of the transferrin receptor (TfrR) (48).

In our FRT cell line that stably expressed KCa3.1, knockdown of exocyst subunits Sec3, Sec6, or Sec8 by siRNA transfection significantly reduced both KCa3.1 current (Figs. 2 and 4) and the basolateral population of KCa3.1 in FRT epithelia (Figs. 1 and 4). These results suggest that fewer postsecretory vesicles containing KCa3.1 channels were able to integrate into the basolateral membrane. Interestingly, the knockdown of a single exocyst subunit was sufficient to disrupt KCa3.1 trafficking, suggesting that the whole eight-protein complex is required for the correct tethering rather than delivery of KCa3.1 to the basolateral membrane. It should be noted that during our immunoblot studies, we specifically probed for streptavidin-labeled, i.e., basolateral KCa3.1. Thus, we specifically quantified basolateral membrane KCa3.1 expression rather than total expression of KCa3.1. Given this, we cannot directly assess whether the KCa3.1 channels that fail to traffic to the basolateral membrane remain in an intracellular compartment or are targeted for lysosomal or proteasomal degradation. In this regard, we previously demonstrated that KCa3.1 is targeted to both the lysosomes and proteasome for degradation and this is enhanced by mutations in several intracellular domains (19, 42). Parenthetically, Nakayama and coworkers reported that in HeLa cells where Sec6 or Sec8 were knocked down this led to an accumulation of vesicles containing TfrR (48). Nonetheless, what is important is the basolateral membrane expression of KCa3.1. Activation of these channels results in hypopolarization of the cell increasing the driving force for ion and water transport function. Therefore, the role of the exocyst proteins in the tethering of KCa3.1-containing vesicles is critical for the membrane expression of KCa3.1 and cellular function.

We also investigated possible protein-protein associations between Sec6/Sec8 and KCa3.1 using a Co-IP assay. Surprisingly, Sec6, but not Sec8, Co-IPed with KCa3.1 (Fig. 5). These data may suggest that associations between exocyst and KCa3.1 are subunit-specific. It is possible that there is still an association between Sec8 and KCa3.1 but this association is relatively weak, owing to the fact, that tethering is both transient and reversible (27).

Given that Devor and coworkers have also demonstrated that trafficking of the genetically related KCa2.3 is also dependent on SNAP23 and STX-4 in HeLa cells (49), exocyst may regulate membrane tethering of other members of the KCNN gene family into the plasma membrane, including KCa2.3. Further experimental studies are required to determine if exocyst is required for the tethering of KCa2.3-containing vesicles to the plasma membrane.

Given the critical role we demonstrate for Sec3, Sec 6, and Sec8 in the basolateral targeting of KCa3.1, it is important to consider the role of these proteins in the exocyst complex as a whole. In this regard, Guo and coworkers (54), using cryo-EM and chemical cross-linking MS, defined the structure of the exocyst complex. The authors described that the eight components of exocyst intricately interweave with each other via intertwined long coiled coils. Indeed, interactions are divided as four pairs: Sec3–Sec5, Sec6–Sec8 that comprise Subcomplex I; while Sec10–Sec15, and Exo70–Exo84 assemble as Subcomplex II of the exocyst complex. In further detail, with each pair of subunits, the coiled-coil regions participate in pairwise associations establishing an intact antiparallel “zipper” which Guo and coworkers (54) have termed the CorEx. In addition, the four dimeric pairs assemble into larger structures, by which the CorEx motifs of Sec3–Sec5 and Sec6 and Sec8 form the four-helix bundle of the tetrameric Subcomplex I. The similar associations occur with Sec10–Sec15 and Exo70–Exo84 to form the tetrameric subcomplex II (54).

Ultimately, additional associations and interactions result in the holo-exocyst complex as two layers. Sec3, Sec6, and Sec8 of subcomplex I play roles in the stabilization of the whole complex (54). In subcomplex I, Sec3 has a domain that inserts into a cleft of Sec8, a portion of Sec5 protrudes into a loop of Sec8; whereas Sec5 clings to a domain of Sec6 all leading to stabilization of this subcomplex (54). In addition, Sec8 bridges to multiple interaction interfaces with Sec10, Sec 15, and Exo84 of subcomplex II. Finally, domains of Sec6 are thought to interact with the four-helix bundle of subcomplex II to stabilize the conformation of Sec6 in the exocyst complex (54). These interactions and associations of the subunits of ecxocyst are quite dynamic and complicated. We direct the reader to Guo and coworkers (54) for additional details of the structural aspects of the hierarchical assembly of the exocyst complex.

As shown in the present study, that knockdown of Sec3, Sec6, and Sec8 results in reduced KCa3.1 in the basolateral membrane population and functional expression of KCa3.1 in epithelial cells. Given the critical role of these exocyst proteins detailed earlier, the most parsimonious explanation for our results is that knockdown of the exocyst subunits results in a destabilization of the exocyst complex and hence reduced trafficking of KCa3.1 to the basolateral membrane.

Based on the data presented herein, we propose the following model of exocyst-mediated trafficking of KCa3.1 (shown schematically in Fig. 7). Trafficking of KCa3.1-containing vesicles to the basolateral membrane is dependent on the Rab-GTPase Rab8. Previously Wu et al. (28) identified that the C-terminal domain of the exocyst subunit Sec15 interacts with several Rab-GTPases including Rab8 and Rab11 using GST-pulldowns. These authors suggest that the binding of Rab proteins to Sec15 is required for the recruitment of exocyst. Similarly, Rab8 may mediate the recruitment of exocyst to KCa3.1-containing vesicles. Contrary to this, Fölsch et al. (65) using an immunofluorescence approach, demonstrated that exocyst recruitment was dependent on the µ1B subunit of AP-1B. Since the trafficking of KCa3.1 is independent of µ1B (21), this raises the question of whether Rab8 alone is sufficient to recruit exocyst to KCa3.1-containing vesicles.

Figure 7.

Proposed cell model of exocyst-mediated intermediate-conductance calcium-activated potassium channel (KCa3.1) trafficking in polarized epithelia. Exocyst is recruited to KCa3.1-containing vesicles, possibly mediated by Rab8. The exocyst subunits Sec3 and Exo70 likely bind to the phospholipid phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂] in the plasma membrane. Exocyst then initiates Soluble N-ethylmaleimide-sensitive factor (SNF) Attachment Receptors (SNARE)-mediated fusion. The SNARE proteins VAMP3, syntaxin-4 (STX-4), and SNAP23 form a SNARE complex resulting in the fusion of the secretory vesicle and the basolateral membrane (not shown for clarity). This results in KCa3.1 being embedded in the basolateral membrane. Figure not to scale. Created with https://app.biorender.com. GTP, guanosine triphosphate; TGN, trans-Golgi network; (for other abbreviations, see main text).

Furthermore, interactions between exocyst and the cargo-binding domain of myosin-V motor protein Myo2 have been reported in S. cerevisiae (66, 67). We have previously demonstrated that delivery of KCa3.1 to the basolateral membrane is dependent on the cytoskeleton and the motor protein myosin-Vc, a mammalian equivalent of yeast Myo2 (22). It is therefore possible that exocyst may be key to the interaction between secretory vesicles and elements of the cytoskeleton. However, experimental evidence is lacking in mammalian in vitro models. Once the vesicle containing KCa3.1 is in close proximity to the basolateral membrane, the PH domains located in the N-terminal region of Sec3 and basic residues in the C-terminal domain of Exo70 bind to PI(4,5)P₂ embedded in the basolateral membrane (30, 31, 68, 69). This interaction tethers the secretory vesicle with the plasma membrane. Then, the proximity of the KCa3.1-containing vesicle and the basolateral membrane enables exocyst to initiate SNARE-mediated fusion. As stated earlier, we have recently reported that the SNARE proteins SNAP23, VAMP3, and STX-4 are required for the insertion of KCa3.1 into the basolateral membrane (23). However, we do know that knockdown of Sec6 or Sec8 does not affect the cellular levels of SNAP23, VAMP3, or STX-4 (Fig. 6) during the tethering of KCa3.1-containing vesicles. Further study is required to determine the sequential molecular events of the exocyst-SNARE mechanism for vesicle incorporation into the basolateral membrane.

Altogether, we have provided the first evidence of a member of the KCNN gene family being regulated by the exocyst complex. Our Ussing chamber and immunoblot experiments demonstrate that when subunits of the exocyst complex were transiently knocked down, this significantly reduced the basolateral population and functional expression of KCa3.1. These data suggest, combined with our protein association experiments, that the exocyst complex regulates the tethering of KCa3.1-containing vesicles to the basolateral membrane before SNARE-dependent insertion of channels into the basolateral membrane of epithelial cells.

DATA AVAILABILITY

Data will be made available upon reasonable request.

GRANTS

K.L.H. was the recipient of an AIM Grant from the Department of Physiology of the UoO. D.C.D. was supported by Cystic Fibrosis Foundation under Grant Number DEVOR22GO and the National Institutes of Health under Grant Number HL092157. R.E.F. was supported by an AIMs Grant from the Department of Physiology. F.J.M. and T.T.C. were supported by a New Zealand Lottery Health Board Grant R-LHR-2019-101706. M.J.E.L. was supported by a doctoral scholarship and a postgraduate publishing bursary both from the UoO.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

T.T.C., F.J.M., and K.L.H. conceived and designed research; M.J.E.L., R.E.F., Y.E-B., and T.T.C. performed experiments; M.J.E.L., R.E.F., F.J.M., and K.L.H. analyzed data; M.J.E.L., Y.E-B., and K.L.H. interpreted results of experiments; M.J.E.L. prepared figures; M.J.E.L. drafted manuscript; M.J.E.L., R.E.F., Y.E-B., T.T.C., D.C.D., F.J.M., and K.L.H. edited and revised manuscript; M.J.E.L., R.E.F., Y.E-B., T.T.C., D.C.D., F.J.M., and K.L.H. approved final version of manuscript.